Presentation is loading. Please wait.

Presentation is loading. Please wait.

VHDL Synthesis in FPGA By Zhonghai Shi February 24, 1998 School of EECS, Ohio University.

Similar presentations


Presentation on theme: "VHDL Synthesis in FPGA By Zhonghai Shi February 24, 1998 School of EECS, Ohio University."— Presentation transcript:

1 VHDL Synthesis in FPGA By Zhonghai Shi February 24, 1998 School of EECS, Ohio University

2 Agenda Getting Started with VHDL VHDL Coding Hint VHDL Coding in FPGAs Floorplanning the Design Building Design Hierarchy Understanding High Density Design Flow

3 Getting Started with VHDL VHDL? Understanding HDL Design Flow for FPGAs

4

5 Getting Started with VHDL Advantages of Using HDLs to Design FPGAs –Top-Down Approach for Large Projects –Functional Simulation Early in the Design Flow –Automatic Conversion of HDL Code to Gates –Type Checking –Early Testing of Various Design implementations

6 Getting Started with VHDL Software Requirements

7 HDL Coding Hints Comparing Synthesis and Simulation Results –Omit the Wait for XX ns Statement wait for XX ns; –Omit the...After XX ns Statement Q <=0 after XX ns;

8 HDL Coding Hints –Order and Group Arithmetic Functions ADD <= A1 + A2 + A3 + A4; ADD <= (A1 + A2) + (A3 + A4); –Omit Initial Values variable SUM: INTEGER :=0;

9 HDL Coding Hints Selecting VHDL Coding Styles –Selecting a Capitalization Style –Using Labels –Using Named and Positional Association –Creating Readable Code –Using Std_logic Data Type

10 HDL Coding Hints Using Schematic Design Hints with HDL Designs –VHDL Design of Barrel Shifter Design implemented using 16 16-to-1 multiplexers, one for each output. –20-input function requires at least 5 logic blocks –16 x 5 = 80 logic blocks implemented using 32 4-to-1 multiplexers arranged in two levels of sixteen. –32 x 1 = 32 logic blocks

11 HDL Coding Hints Resource Sharing & Gate Reduction

12 HDL Coding Hints Using If Statements, Using Nested_If Statements Using Case Statements

13 HDL Coding Hints

14

15

16

17 HDL Coding for FPGAs latches vs. flip-flop in Xilinx FPGAs: –2 FFs/CLB which can be used –latches requires FG function generator using latch... if (write = ‘1’ ) then stored_value <= value_in; end if;...

18 HDL Coding for FPGAs Latches vs. flip-flops using flip-flop... if (write’event and write = ‘1’) then stored_value <= value_in; end if;...

19 HDL Coding for FPGAs Encoding State Machines –Using Binary Encoding –Using Enumerated Type Encoding –Using One-Hot Encoding Better suited for use with the fan-in limited and flip-flop-rich architecture of FPGA

20

21 HDL Coding for FPGAs Comparing Synthesis Results for Encoding Styles

22 HDL Coding for FPGAs Implementing Multiplexers with Tristate Buffers –Use internal tristate buffers (BUFTs) to implement multiplexers larger than 4-to-1. Can vary in width with only minimal impact on area and delay Can have as many inputs as there are tristate buffers per horizontal longline in the target device Have one-hot encoded selector inputs

23

24

25 HDL Coding for FPGAs

26 Floorplanning Your Design Using the Floorplanner –Creating a MAP File Using Xmake Using PPR Using Prep for Floorplanner Command

27 Floorplanning Your Design –Deciding What Elements to Floorplan Large objects such as RPMs, registers, counters, and RAMs Buses (place all BUFTs and bus elements) BUFTs with I/O or RPM inputs Multiple BUFTs (except VCC or GND) with identical source pin inputs

28 Building Design Hierarchy –Advantages Efficiently manage the design flow Reduces design time by allowing you to use existing design modules more than once Produce designs that are easy to understand

29 Building Design Hierarchy Modifying Design Hierarchy for PPR –Reduces Gate Count –Improves Routability –Reduces Routing Time –Reduces Time Required for Small Design Changes –Reduces Debugging Time

30 Building Design Hierarchy Top Design Example

31 Compiling Top Design as One Flat Module

32 Building Design Hierarchy Compiling Top Design After Modifying the Hierarchy –R0 block uses approximately 591 CLBs –X0 block uses approximately 342 CLBs –UP0 block uses approximately 25 CLBs –DD0 block uses approximately four CLBs

33 Building Design Hierarchy Compiling Top Design After Modifying the Hierarchy

34

35

36 Building Design Hierarchy

37 Comparing Top Design Methodologies –Flat Design densely packed and is unroutable. –Original Design Hierarchy small changes to this design may make the design unroutable. –Modified Hierarchy

38

39 Understanding High-Density Design Flow

40 Estimating Your Design Size –run PPR on your design after compiling it as one flat module –Determining Device Utilization Evaluating Your Design for Coding Style and System Features –correct coding style problems –incorporate FPGA system features

41 Understanding High-Density Design Flow Modifying Your Design Hierarchy –One flat module vs. many small modules –structure design hierarchy to guide the placement and routing. Synthesizing and Optimizing Your Design –Use the Synopsys Group command to define the new hierarchy.

42 Understanding High-Density Design Flow Use the Synopsys Group command {M1,M2,M3,M4,M5} -design_name X1 -cell_name X1 group {M6,M7,M8,M9,M10,M11} -design_name X2 \ - cell_name X2 group {N7,N8,N9,N10,N11,N12,N13} -design_name R1 \ -cell_name R1 group {N2,N3,N4,N5} -design_name R2 -cell_name R2 group {N6} -design_name R3 -cell_name R3 group {N1} -design_name R4 -cell_name R4

43 Understanding High-Density Design Flow Translating Your Design and Adding Group TimeSpecs –translate your design to an XNF file –adding Timing Specifications Building Your Design Hierarchy –constrain your design modules to specific device areas in the Floorplanner. –define boundaries in the Floorplan window and place the selected modules within the specified boundaries.

44 Understanding High-Density Design Flow Floorplanning Your Design Placing and Routing Your Design –Using PPR Options –Determining If PPR Can Route Your Design Evaluating the Results –Evaluating Module Placement with the Floorplanner

45

46

47

48


Download ppt "VHDL Synthesis in FPGA By Zhonghai Shi February 24, 1998 School of EECS, Ohio University."

Similar presentations


Ads by Google